Thanks Jan, we added several additional checks for non-standard protocol responses, and also made the client revert to DNS seeding more quickly if it runs into trouble, so it's now more robust against sybil/DOS attack. I mentioned in the coindesk article that I didn't think what your nodes were doing was intended to be malicious with respect to network disruption. It's our job to better handle non-standard or even malicious behavior from random p2p nodes.


Aaron Voisine
co-founder and CEO
breadwallet.com

On Mon, Mar 16, 2015 at 1:44 AM, Jan Møller <jan.moller@gmail.com> wrote:
What we were trying to achieve was determining the flow of funds between countries by figuring out which country a transaction originates from. 
To do that with a certain accuracy you need many nodes. We chose a class C IP range as we knew that bitcoin core and others only connect to one node in any class C IP range. We were not aware that breadwallet didn't follow this practice. Breadwallet risked getting tar-pitted, but that was not our intention and we are sorry about that.

Our nodes DID respond with valid blocks and merkle-blocks and allowed everyone connecting to track the blockchain. We did however not relay transactions. The 'service' bit in the version message is not meant for telling whether or how the node relays transactions, it tells whether you can ask for block headers only or full blocks.

Many implementations enforce non standard rules for handling transactions; some nodes ignore transactions with address reuse, some nodes happily forward double spends, and some nodes forward neither blocks not transactions. We did blocks but not transactions.

In hindsight we should have done two things:
1. relay transactions
2. advertise address from 'foreign' nodes

Both would have fixed the problems that breadwallet experienced. My understanding is that breadwallet now has the same 'class C' rule as bitcoind, which would also fix it.

Getting back on the topic of this thread and whether it is illegal, your guess is as good as mine. I don't think it is illegal to log incoming connections and make statistical analysis on it. That would more or less incriminate anyone who runs a web-server and looks into the access log.
At lease one Bitcoin service has been collecting IP addresses for years and given them to anyone visiting their web-site (you know who) and I believe that this practise is very wrong. We have no intention of giving IP addresses away to anyone, but we believe that you are free to make statistics on connection logs when nodes connect to you.

On a side note: When you make many connections to the network you see lots of strange nodes and suspicious patterns. You can be certain that we were not the only ones connected to many nodes.

My takeaway from this: If nodes that do not relay transactions is a problem then there is stuff to fix.

/Jan

On Fri, Mar 13, 2015 at 10:48 PM, Mike Hearn <mike@plan99.net> wrote:
That would be rather new and tricky legal territory. 

But even putting the legal issues to one side, there are definitional issues.

For instance if the Chainalysis nodes started following the protocol specs better and became just regular nodes that happen to keep logs, would that still be a violation? If so, what about blockchain.info? It'd be shooting ourselves in the foot to try and forbid block explorers given how useful they are.

If someone non-maliciously runs some nodes with debug logging turned on, and makes full system backups every night, and keeps those backups for years, are they in violation of whatever pseudo-law is involved?

I think it's a bit early to think about these things right now. Michael Grønager and Jan Møller have been Bitcoin hackers for a long time. I'd be interested to know their thoughts on all of this.

------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Bitcoin-development mailing list
Bitcoin-development@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bitcoin-development



------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Bitcoin-development mailing list
Bitcoin-development@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bitcoin-development